Download The invisible hand of grammaticalization

The invisible hand of grammaticalization
West-Germanic substitutive infinitive and the prefix ge-*
Livio Gaeta
Università di Napoli “Federico II”
Grammaticalization may have therapeutic and pathological effects on
morphology. The paper will focus on these latter with special regard to the
occurrence of a morpheme in an unexpected form as is the case for the WestGermanic substitutive infinitive. The reason for this mysterious case of formmeaning mismatch must be sought in the grammaticalization of the Germanic telic
prefix *ga-. As a consequence of its grammaticalization in the past participle, a
semantic incompatibility prevented the so-called AcI-verbs from being touched
by the grammaticalization wave spreading the perfect periphrasis throughout
the whole verbal system. Thus, the arguably default form came in, namely the
infinitive, whereby the perfect periphrasis could be completely paradigmaticized
even though at the cost of a form/meaning mismatch. In this light, the long-wave
effect of grammaticalization can be made responsible for the anomaly preserved
until today in all West-Germanic dialects, in which ge- was grammaticalized as an
inflectional marker.
1. Introduction
What is the impact of grammaticalization on morphology? Basically, a therapeutic
one: Grammaticalization is generally responsible for the increase of iconicity (cf.
Bybee 2006). For instance, in Old English, the same form full could be used both
as an adjective and as an adverb. After the grammaticalization of līc ‘body’ as the
adverb-forming suffix -ly, iconicity was increased because of the actual explicit
coding of the derivational relation between the basic adjective full and the derived
adverb fully.
*Parts of this paper were also presented at the 39th SLE Conference held in Bremen, August
2006. I am deeply indebted with the people present on these occasions for insightful remarks and
criticism, and especially Elke Ronneberger-Sibold and Pieter Seuren. I am also indebted with the
editors of the volume and an anonymous reviewer for suggestions and remarks. Needless to say, I
am solely responsible for errors and misunderstandings contained in the paper.
 Livio Gaeta
On the other hand, pathological effects can be observed: Dysfunctional morphological structure comes about which violates general and/or language-specific
morphological principles (cf. Gaeta 2008). In other words, grammaticalization
may give rise to mismatches, which can be defined as “synchronic form/meaning discrepanc[ies], in which form-function mappings are incongruent with
respect to more general patterns of correspondence in the language” (Francis &
Michaelis 2003: 2).
In the rest of the paper, several different cases of form/meaning mismatch
will be discussed. In particular, the so-called substitutive infinitive occurring in
German as well as in other West-Germanic languages will be focused on, with
the aim of showing that this quirky feature may be explained as a side-effect of
the grammaticalization of the perfect periphrasis in the Middle Ages. The paper is
structured as follows: Section 2 provides a survey of different cases of form/meaning mismatches which can be explained as the result of grammaticalization, while
Section 3 introduces the historical background which is necessary for understanding the rise of the substitutive infinitive. The latter will be explained in Section 4 as
the peculiar expansion of the perfect periphrasis to a particular verb class, namely
to verbs governing a bare infinitive. The final Section 5 draws the conclusion.
2. Form/meaning mismatches in morphology
A first type of form/meaning mismatch is given by the appearance of a certain
form in the “wrong” place. “Trapped” morphology comes about as a consequence
of grammaticalization (cf. Harris & Faarlund 2006). Accordingly, a morpheme
turns out to be “trapped” by the grammaticalization of another morpheme. For
instance, in Old Norse the inflectional marker -s was “trapped” in a dispreferred
position and subsequently eliminated in Norwegian:
(1) ON hest-s = in-s
>
Norw. hest-en-s
horse-gen = the-gen
“the horse (gen.)”
Similarly, introflection came about in Latin as a consequence of the grammaticalization of markers like -pse and -dam (cf. Haspelmath 1993):
(2) a. +is-pse/ea-pse
‘self ’ > ipse, -a, -um
eum-pse/+eam-pse
> ips-um, -am
+eō-pse/eā-pse
> ips-ō, -ā
b.
nom. quī-dam/quae-dam ‘a certain one’
gen. cuius-dam
dat. cuī-dam
acc. quen-dam/quan-dam
abl. quō-dam/quā-dam
The invisible hand of grammaticalization 
Notice that only in the first case a morphological change, called externalization of
inflection by Haspelmath, has eliminated introflection, while in the second case
the mismatch was preserved.
A second type of form/meaning mismatch is given by the appearance of
an unexpected form for a given bundle of morphosyntactic features. This usually comes about as a result of the expansion/paradigmaticization of a certain
grammaticalized construction. In some Russian dialects, the grammaticalization
of the perfective periphrasis has given rise to a sort of split alignment, because
the possessive construction which is at the origin of the perfective periphrasis
contained a non-nominative subject (cf. Creissels 2008: 25):
(3) a.
U Ivana
novaja
mašina.
at Ivan:gen new:sg:f:nom car: sg:nom
“Ivan has a new car.”
b. U volkov
s’’edeno
korovu.
at wolf:pl:gen eaten:sg:n cow:sg:acc
“The wolves have eaten the cow.”
On this background, the rest of the paper will be devoted to discussing a particulary intriguing case of synchronic mismatch, namely the so-called substitutive
infinitive ( = IPP, Infinitivus Pro Participio) occurring in several West-Germanic
languages like Modern Standard German ( = MSG), Dutch, West Flemish, Afrikaans (cf. Schmid 2005), and exemplified below by MSG:1
(4) a.
Ulli hätte die Morphologie-Tagung nicht besser organisieren können.
Ulli had the morphologyˉmeeting not better organize:inf can:inf
“Ulli couldn’t organize better the morphology conference.”
b. *Ulli hätte die Morphologie-Tagung
Ulli had the morphology meeting
organisieren gekonnt.
organize:inf can:pstptcp
c.
Ulli hätte es gekonnt.
Ulli has it can:pstptcp
“Ulli could do it.”
nicht besser
not better
Similar to the Russian dialects mentioned above, the mismatch is given here by the
appearance of an unexpected infinitive for what should have been a past participle.
Notice that the latter is indeed attested, when the modal is used as a full verb, as
shown in (4c).
1. Since a lot of individual and regional variation occurs, the label MSG refers to the standard
variety as usually described by reference grammars like for instance DUDEN (2005: 473).
 Livio Gaeta
Besides modals, the IPP occurs in MSG optionally with causative and perception
verbs:
(5) a.
Ulli hat den Wagen kommen sehen/gesehen.
Ulli has the car
come:inf see:inf/see:pstptcp
“Ulli has seen the car coming.”
b. Ulli hat den Wagen kommen lassen/gelassen.
Ulli has the car
come:inf let:inf/let:pstptcp
“Ulli has got the car to come.”
Several explanations have been suggested for understanding the IPP and more
generally the character of the West-Germanic verbal complex (cf. Gaeta 2005
and Wurmbrand 2006 for overviews of respectively diachronic and synchronic
approaches); however, a satisfactory “answer to the question of what the (deep)
motivation of verb cluster formation is and why this phenomenon only exists in
certain languages is still outstanding” (Wurmbrand 2006: 234).
In what follows, it will be shown that far from being a quirky character of the
West-Germanic verb complex, this mismatch has to be seen as a side-effect of the grammaticalization of the perfect in the late Middle Ages, in consequence of the special role
played by the telic prefix ge- of the past participle.
3. The ge-prefix and the perfect periphrasis
Two important facts can be observed at the Jahrtausendwende in Continental
West-Germanic:2 the grammaticalization of the ge-prefix as an inflectional marker
in the past participle, and the grammaticalization of the perfect periphrasis. Both
facts are fairly complex and call into play different properties as well as domains.
The prefix ge- played a crucial role in the history of the West-Germanic languages: in Proto-Germanic, it is generally reconstructed as a proclitic marker of
telicity (or “perfectiveness” for Lowe 1972: 223), as shown by forms like Gothic
ga-nisan ‘to be cured’, being related to Latin cum, Old Irish com-, Albanian kë, etc.
In the West-Germanic languages, the prefix ge- is ever since attested as an additive
marker of the past participle forming a sort of circumfix, as it is still the case in
MSG geschlafen and Dutch geslapen ‘slept (pstptcp.)”. However, its telic value was
2. In what follows, Old English is intentionally left aside: although similar phenomena are
attested, the partially different conditions and the strong Romance influence after the battle
of Hastings render the case of Old English much more complex than what can be dealt with
here. See at least Wischer & Habermann (2004) for the prefix ge- in Old English, and Fischer
(1994) for the development of the infinitival complementation.
The invisible hand of grammaticalization 
not lost, at least in the earlier stages. In fact, in Old High German ( = OHG) the
prefix was not compatible with intrinsically telic verbs like the following ones (cf.
Braune/Reiffenstein 2004: 273):
(6)
bringan ‘bring’
findan ‘find’
quëman ‘come’
trëffan ‘hit’
wërdan ‘become’
→ brungan/brāht ‘(pstptcp.)’
→ fundan ‘(pstptcp.)’
→ quëman/quoman ‘(pstptcp.)’
→ troffan/(gitroffan) ‘(pstptcp.)’
→ wortan/(giwortan) ‘(pstptcp.)’
This does not surprise given the intrinsic telic value of the prefix, which makes it
highly redundant with telic verbal bases. In other words, the grammaticalization process
leading to the expansion of ge- throughout the entire verbal system as observed in
MSG was not yet completed. Notice that in Middle High German ( = MHG) this
restriction still holds (cf. Paul et al. 2007: 247):
(7) MHG brâht
MSG
funden
komen
troffen
worden
gebracht
gefunden
gekommen
getroffen
(ge)worden
The grammaticalization of the prefix ge- as an inflectional marker was not only
incomplete, thus giving rise to inconsistencies across the verb paradigms, but also
accompanied by the phenomenon of layering, namely “the persistence of older
forms and meanings alongside newer forms and meanings, whether derived by
divergence from the same source or by renewal from different sources” (cf. Hopper &
Traugott 2003: 124). In fact, ge- could be generally prefixed to verbs giving rise to
aspectually distinct pairs, as shown below with the help of AcI-constructions:
(8) OHG sehan ‘see’/gisehan:
a.
thar sah si drúhtin stantan
there saw she lord
stand:inf
“She saw the lord standing there.”
(Otfr., Ev. 5.7. 44)
b.
thar gisah er stantan gotes boton
sconan
there gi-saw he stay:inf god’s messenger beautiful
“He came to see god’s beautiful messenger standing there.”
(Otfr.,ˉEv.ˉ1.4.21)
OHG hōren ‘hear’/gihōren:
c.
then
fater hort er sprechan
the:acc father heard he speak:inf
“He heard the father speaking.”
d. ik gihorte dat seggen … (Hld. 1)
I gi-heard this say:inf
“I came to hear saying this, ...”
(Otfr., Ev. 1.25.15)
 Livio Gaeta
Notice that the usage of the prefixed companion with the AcI-construction forces
an ingressive interpretation in (8b) (cf. Schrodt 2004: 125).3 As for the example of
gihōren in (8d) above, the verse is the very beginning of the Hildebrandlied.
It must be emphasized that AcI-verbs constituted a particular verb class,
which required as a necessary condition for the AcI-construction to be possible
the simultaneity between the two events conveyed by the predicates. In other
words, the two events had to be interpreted as at least partially overlapping.4
The class mainly contained perception and causative verbs; however, an extension
to other verb classes (saying, thinking verbs, etc.) took place quite soon under
Latin influence (cf. Speyer 2001). In this light, the usage of the aspectually marked
verbs prefixed by gi- was only possible on condition that the first event had not
to be interpreted as closed with respect to the second one. Accordingly, the pairs
reported in (8) above display an ingressive meaning.
This prefixal usage of ge- is still highly productive in MHG, as can be gathered
from the following table (cf. Solms 1991: 125), which shows the distribution of
certain verbal prefixes in OHG and MHG texts:
Table 1. Distribution of verbal prefixes in OHG and MHG
ge-
ent-
er-
be-
ver-
zer-
OHG Abrogans
28,3%
4,3%
19,2%
27,4%
15,5%
5,3%
MHG Iwein MS A
MHG Gregorius MS A
33,8%
32,1%
6,3%
7,3%
18,8%
19,4%
18,8%
21,9%
21,3%
20,8%
1,1%
2,2%
In particular, the prefixed companion was used to convey future and pluperfect
meaning still in MHG:
3. Schrodt (2004: 113) distinguishes between an ingressive and an egressive aspect: they
both realize the “complexive” dimension, which generally represents a bounded futural event.
The opposition between the prefixed and the prefixless verb allows one to distinguish “die
wichtigeren Begebenheiten im Bericht von den weniger wichtigen ..., die ihrerseits in der konstatierenden Aspektfunktion mittels des Simplex erscheinen” (Schrodt 2004: 126). I will not
go further into the rather complex picture depicted by Schrodt, also because I am aware of the
large amount of theoretical and terminological differences to be found in studies on tense and
aspect. I will limit myself to speaking of telicity (or boundedness) as for the actional/aspectual
dimension, and of anteriority/posteriority for the temporal dimension.
4. Notice that this condition still holds for MSG:
Den Vater hörte er sprechenˉ/ ˉ*gesprochen haben.
the father heard he speak/ spoken
have:inf
The invisible hand of grammaticalization 
(9) a.
Future
(cf. Paul et al. 2007: 289):
ich weiz wol waz Kriemhilt mit disem scatze getuot
I know well what K.
with this treasure ge-does
“I well know what Kriemhilt will do with this treasure.”
b. Pluperfect
(cf. Paul et al. 2007: 290):
als
der künic Gunther die rede
wol gesprach, Hagene
when the king G.
the discourse well ge-spoke, H.
der küene den guoten Rüedegêren sach.
the clever the good R.
saw
“When the king Gunther had finished his discourse, the clever Hagen saw
the good Rüedegêren.”
Thus, the prefix ge- was a key marker of telicity for verbs during the whole Middle
Ages, being layered in different but homogeneous usages.
In the same time span, namely in late OHG/early MHG, the grammaticalization of the perfect periphrasis also took place (cf. Kotin 1997; Harris 2003). In classical OHG, a sentence such as the following one displayed a biclausal structure, in
that the verbal complex had a resultative reading and an agreement marker on the
past participle occurred. Furthermore, it was only possible with transitive verbs,
which allowed for a possessive reading:
(10)
iogiuuelich, thie thar gisihit uuib
sie zi geronne, iu habet sia
whoever who there gi-sees woman her to lust
he has her
forlegana in sinemo herzen
enclosed in his
heart
(Tat., Ev.Harm. 28.1)
“Whoever will see a woman in order to lust after her, he’ll have already got her
enclosed in his heart.”
In late OHG, beginning with the 11th century, the same construction may be considered grammaticalized, thus forming a monoclausal sentence with the addition
of temporal reference, namely anteriority:5
(11) ˉa. dáz chît
síben bûohlísto.
this means seven arts
dîe
unmánige gelírnet hábent.
which un-many learned have
únde áber mánige genémnen chûnnen
and but many gi-name:inf can
(Not.ˉI.65.3–4)
“This means the seven arts, which many have studied, but only few are able to
mention.”
5. In more formal terms, we can say that the grammaticalized auxiliary haben provides a
further temporal interval which allows the past participle to release its referential potential.
Notice that in the case of the AcI-construction, the second temporal interval of the perception
or causative verb, i.e. its second argument-time window, is saturated by the embedded verb
(cf. Klein 2000; Gaeta 2008 for details).
 Livio Gaeta
b. dînen zórn fúrhtendo hábo íh … geuuéinot
your anger fearing
have I
cried
“Fearing your anger I have cried.”
(Not. 3.14.10, Ps. 6)
Furthermore, we observe the complete loss of any agreement on the participle,
and, more significantly, the expansion to other verb classes, and in particular to
intransitives (cf. Kotin 2000: 332), as shown by (12b).
4. The IPP and the expansion of the perfect periphrasis in MHG
What is the place of the IPP in the necessarily sketchy picture depicted so far?
Before looking at the historical evidence avalaible, let us speculate on the role
which has to be attributed to the IPP in this scenario.
The main thesis which I would like to defend is that the IPP has to be seen
as a side-effect of the grammaticalization of the perfect in the late Middle Ages, a
consequence of its expansion to all possible verb sets. In this light, the mismatch
is motivated by the semantic incompatibility of the past participle with the AcIconstruction. The incompatibility is due to the special role played by the prefix
ge- of the past participle, which, as we have seen, is intrinsically telic (for a partially
similar view, see also Ijbema 1997: 152). Therefore, it provides an obstacle for the
expansion of the perfect periphrasis to the AcI-construction, because the latter
requires a partial overlapping of the two events encoded. The crucial point is that the
perfect periphrasis as such is theoretically compatible with the AcI-verbs, because it
conveys temporal reference of anteriority, which can be predicated of the first event
of the AcI-construction. The problem is, however, that the past participle is also
marked as telic by the prefix ge-, which is incompatible with the AcI-construction,
unless an ingressive reading is forced, as we have seen above.6 The latter is in its
6. An anonymous reviewer questions whether the semantic incompatibility of the geprefixed participle should be the right explanation here in the light of the acceptability of
sentences like Karl sah ihn geschlagen ‘Karl saw him knocked down’, which contains a past
participle depending on a perception verb. However, such structures cannot be assimilated to
AcI-constructions because no infinitive occurs, which is crucially specified for simultaneity
with the perception verb. In fact, the telicity effect only arises when the participle is forced
to govern an embedded infinitive specified for unboundedness. When no simultaneity is
required, the past participle shows up instead of the IPP: Karl hat ihn geschlagen gesehen/
*sehen ‘Karl has seen him knocked down’. Notice that such constructions can be assimilated to
those in which a past participle modifies a noun, which are impossible with unergative verbs: der
geschlagene Riese ‘the giant knocked down’ vs. *der geschlafene Riese ‘lit., the slept giant’. As
The invisible hand of grammaticalization 
turn incompatible with the perfect periphrasis. To overcome the obstacle represented by the prefixed participle, the infinitive, as a default form (cf. Eisenberg,
Smith & Teuber 2001: 256), replaced the participle, thus allowing the extension of
the perfect to the AcI-verbs. The construction did convey anteriority of the first
event with respect to the speech time, while at the same time the infinitive was not
intrinsically marked for boundedness.
In spite of the apparently speculative character of this hypothesis, a number of
tight predictions logically result from it. Firstly, the IPP should appear concomitant
with the extension of the perfect construction to complex verb structures. Thus,
we wouldn’t expect the IPP to appear immediately after the grammaticalization of
the perfect periphrasis. Instead, because of the obstacle represented by the prefixed
participle, a delayed expansion is predicted. Secondly, the complex verb structures
affected earlier by the IPP should be well characterized as for their aspectual/
actional composition. This would reveal the role played by the aspectual/actional
properties of the involved predicates. Finally, the infinitive should be used as a
default form in other verbal periphrases as well in the same time span. Let us see
if these predictions are supported by the historical evidence.
As to the first prediction, the IPP is first attested in the 13th century, two centuries after the grammaticalization of the perfect periphrasis:
(12) a.
hâst1
du daz schif lâzen2 gân3 have:2 sg you the boat let:inf go:inf
“You left the boat go away.”
b. daz er die brieffe nit habe1 ˉheissen2ˉ machin3
that he the letters not have order:inf make:inf
“that he did not order to prepare the letters”
(Trist. 6796–7)
(Publ.ˉ615.31)
These examples, and the others collected by Kurrelmeyer’s (1910) thorough
investigation of the first attestations of the IPP, are interesting from a double
viewpoint. On the one hand, they show that MHG prevalently behaves like
Dutch as for the linearization of the constituents, namely V1V2V3. We will come
back to this point later. On the other, modal verbs, which categorically require
IPP in MSG, were scarcely attested, whereas causative and perception verbs
expected, unergatives are incompatible: *Karl hat ihn geschlafen gesehen ‘lit., Karl has seen him
slept’. Furthermore, unaccusatives pattern with transitives: der vom Mars gekommene Riese ‘the
giant come from Mars’ and Karl hat ihn ins Zimmer gekommen gesehen ‘lit., Karl has seen him
come (pstptc.) into the room’. See Klein (2000) and Gaeta (2008) for more details.
 Livio Gaeta
were robustly attested. The following table summarizes the data collected by
Kurrelmeyer (1910):
Table 2. The earliest attestations of IPP in MHG
13th c.
14th c.
Word order
V1V2V3
V1V3V2
V1V2V3
V1V3V2
causatives
tuon ‘do’
lāŠen ‘let’
heiŠen ‘order’
6
8
10
1
10
4
12 (PP = 12)
1
modals
müeŠen ‘may’
türren ‘dare’
mügen ‘be able to’
wellen ‘want’
künnen ‘understand’
dürfen ‘need’
perception verbs
hœren ‘hear’
sëhen ‘see’
benefactives
hëlfen ‘help’
1
1
1
8 (PP = 1)
11
1 = V3V1V2
(PP = 2)
1
1
(PP = 1)
Although sparse forms of the expected past participles are attested (indicated
by PP in the table),7 the dominant type is given by the IPP, mainly with causative
and perception verbs. Thus, the historical findings project a picture which is very
different from what we observe today: whereas in MSG (and Dutch) modals categorically display the IPP, in its earlier stage the IPP occurred in a systematic way
only with the core set of the AcI-verbs, namely perception and causative verbs.
Far from being unexpected, this fact fits very well into the scenario depicted so
far, and actually fulfills the second prediction formulated above. In fact, modals
do not require the condition on simultaneity which we have observed above for
7. The only exception is given by the significant occurrence of the past participle of heiŠen,
which has been used to support the traditional view that explained the IPP as a consequence
of the confusion between the formally identical infinitive and past participle: heiŠen/geheiŠen
(cf. Grimm 1837: 168). In the light of the robust occurrence of cases in which no identity
occurred like tuon/getān, hœren/gehœrt, this view can be definitely rejected.
The invisible hand of grammaticalization 
AcI-verbs, and as expected were commonly used in MHG in combination with the
perfect periphrasis:8
(13) daz möht ir gerne
hân verdagt
this could you willingly have be-silent:pstptcp
“You could have been silent about this with pleasure.”
(Parz. 464.6)
Thus, the apparently odd distribution of the IPP in MHG can be straightforwardly
explained by the peculiar character of modals which allow for a less restricted
combination of the embedded predicate. In those cases in which the combination
was tighter and required an overlapping of the two events, namely with perception and causative verbs, this escape solution was not available, namely expressing
anteriority on the embedded predicate, and the IPP manifested itself.
As for third prediction formulated above, namely the infinitive as the available
default form, notice that the infinitive was generally being extended in the 13th
and the 14th centuries to other verbal periphrases as well:
(14) a.
tuon + infinitive periphrasis
die lieb sich meren
tuot
the love itself enlarge:inf does
“The love does enlarge itself.”
b.
würde + infinitive periphrasis for the past subjunctive (cf. Ebert et al. 1993: 392):
mit dem Turcken vnd Frantzose dis jar so stehen
wurde
with the Turk
and Frenchman this year so stand:inf would
“This year it would be like so with the Turk and the Frenchman.”
c.
werden + infinitive periphrasis for the future
daz wirt er eine klagen
that will he one lament:inf
“That he will lament one.”
(Paul et al. 2007: 308):
(cf. Paul et al. 2007: 296):
This last usage is particularly significant for our purposes, because the future
periphrasis also came to display the same options of the IPP as for the linearization
8. In this regard, consider Grønvik’s (1986: 46) comment: “Daß Perfektformen von modalen
Hilfsverben im Mhd. nicht auftreten, rührt also daher, daß man die Perfektbedeutung nicht
am Hilfsverb, sondern am Hauptverb ausdrückte” [Thus, that forms of the perfect of modals
did not occur in MHG depends on the fact that the perfect meaning was expressed not by the
modal, but by the main verb]. Again, MSG confirms this insight:
Hans könnte kommen/ gekommen
sein.
Hans could come:inf come:pstptcp be.
‘Hans could have come’
 Livio Gaeta
of the constituents (cf. Kotin 2003 for a detailed reconstruction of the origin of the
werden + infinitive periphrasis):
(15) Ulli glaubt, dass seine Frau ruhig wird
/hat schlafen können.
U. thinks that his wife quietly becomes /has sleep:inf ˉcan:inf
“Ulli thinks that his wife will be able to/could sleep quietly.”
Two further independent arguments support the analysis laid down here: the
global picture of (West-)Germanic languages and the restrictions still displayed by
the past participle in the modern languages.
As for the global picture of (West-)Germanic languages, it must be stressed
that the IPP is only attested in those (West-)Germanic languages, in which gewas grammaticalized as a marker of the participle (cf. Abraham 1998: 109), as
illustrated by two other varieties, namely Frisian and Low German, which do not
display the ge-prefix. Expectedly, the IPP does not show up:
(16)
Frisian:
LowˉGerman:
dat
dat
ˉer it
he dat
boek leze kind
book lesen kunnt
hat
het
that he the book read can:pstptcp has
“that he could read the book”
However, two exceptions have been pointed out which disturb this water-tight picture, namely West Frisian dialects, which display prefixless participles and the IPP,
and Yiddish which is its mirror-image, showing prefixed participles and the IPP:
(17) a.
West Frisian dialects (prefixless participles and IPP):
dat er it hat kinne dwaan
that he it has can:inf do
“that he could do it”
b. Yiddish (prefixed participles and IPP):
Zey hobn gemuzt
zikh
farteydikn
they have must:pstptcp themselves defend
“They had to defend themselves.”
However, it must be noted that strong language contact was at play in both cases,
which might have influenced the reshaping of the verbal complex on the basis of
Dutch (for West-Frisian dialects), or the normalization of the structure (for Yiddish).
Notice that normalized structures are also attested in the diachrony of German,
which corresponds to more general normalization tendencies witnessed by Paul
(1920: 128–129):
“Ausnahmsweise Verwendung des Part. der genannten Verba in der neueren
Sprache ist wohl immer nicht durch landschaftlichen Sprachgebrauch, sondern
durch logische Erwägungen der Schriftsteller veranlaßt [...] Besonders hat
The invisible hand of grammaticalization 
Rückert die Sprache zu korrigieren versucht [...] Auch Adelung hält bei hören das
Part. für die regelmäßigere, edlere Form.9
Thus, it does not come out as a surprise that for such a quirky feature like the IPP
simplification strategies, which are particularly strong in the case of language contact, might have been at work.
Finally, traces of the bounded nature of the German past participle can still be
observed today. In fact, if we try to embed an AcI-governing past participle into a
noun phrase, an ungrammatical result obtains:
(18) a. *Der im
Park schlafen gesehene
Mann wurde verhaftet.
the in:def park sleep:inf see:pstptcp man became arrested
b. L’uomo visto
dormire nel
parco venne arrestato.
the-man see:pstptcp sleep:inf in:def park came arrested
“The man seen sleeping in the park was arrested.”
As testified by the Italian correspondent in (19b), the operation is theoretically
conceivable, and in fact grammatical in a language like Italian in which such a
telicity marker was never present. In MSG, however, traces of the telic nature
of the prefixed past participle still provide an obstacle for the expansion of the
AcI-construction: in fact, the past participle cannot govern an embedded predicate which has to be specified as unbounded without the help of the auxiliary
haben that provides a second temporal interval for the governing predicate to be
anchored at. As we have argued for above, the latter is only possible on condition
that a non-telic form, namely the IPP, shows up.
Lastly, the picture sketched so far would be incomplete without adding a final
word on the question of the deviant linearization. The following table summarizes
the data collected by Schmid (2005) on the linearization of the constituents in the
IPP in five West-Germanic languages when the second constituent is a modal:
Table 3. IPP constituent order in 5 varieties of West-Germanic
modals
MSG
SG
A
WF
B
D
V1V3V2
V1V2V3
V1V3V2V3V1V2
?V2V3V1
V1V2V3
?V2V3V1
V1V2V3
V1V2V3
MSG = German; SG = Sankt Gallen German; A = Afrikaans; WF = West Flemish; B = Bernese German;
D = Dutch
9. [Exceptional employment of the participle of the verbs mentioned above in the modern
language is certainly not always due to regional usage, but to logical remarks of the authors ...
In particular, Rückert attempted at improving the language ... And also Adelung considers the
participle in the case of hören to be the more regular, noble form].
 Livio Gaeta
On the basis of what has been observed above, we may conclude that Dutch
has retained the earlier word order, whereas the other West-Germanic languages
underwent minor changes. At any rate, none of the five varieties displays a complete
head-final linearization. Notice that for MSG the reordering of the constituents
goes back to the 18th century (cf. Härd 1981). To account for the persistence of the
anomaly, it might be hypothesized that the “abnormal” word order was functionally adequate to encode the repair effect of the default-infinitive. In this regard,
recall that a similar deviant linearization is attested for the future periphrasis as well
in MSG, as shown in (16) above. The latter also results from the default extension of
the infinitive in the place of an older construction containing the present participle
(cf. Kotin 2003).
Whatever the reason for the persistence of deviant linearization might be, a
common developmental pattern may be observed. Except for the conservative
order occurring in Dutch, in the other West-Germanic languages a slow process
of reordering has taken place, which aimed at remodeling the verb serialization on
the basis of the dominant verb-final pattern, namely either V2V3V1, in which the
finite verb shifts to the end of the verbal complex, or V1V3V2, in which the nonfinite verbs are inverted consistently with the verb-final pattern. In other words,
we observe a conflict between the tendency towards a uniform verb-final serialization pattern and the functionally useful coding of the anomaly. Notice that this
explanation copes well with the fact that the theoretically conceivable reordering
V2V1V3 does not occur in any West-Germanic variety displaying IPP (cf. Ijbema
1997: 148), because it doesn’t match either of the two strategies suggested above.
Furthermore, a number of grammatical and extra-grammatical factors have
presumably contributed to making the picture intricate, as insightfully commented
by Ebert (1986: 122):
“Die für die Ausdrucksmöglichkeiten des Deutschen relativ unbedeutende
Entwicklung der Stellung des finiten Verbs im Verbalkomplex gehört zu
den schönsten Beispielen des Zusammenwirkens von sprachlichen und
außersprachlichen Faktoren bei der syntaktischen Variation und dem
Syntaxwandel in unserem Zeitraum”.10
At any rate, whether the suggestions regarding the constituent linearization laid
down here are correct or not must be left as an open question for future research,
10. [The development of the position of the finite verb in the verbal complex, which is quite
irrelevant for the expressive possibilities of German, is one of the most beautiful examples of
the interaction of linguistic and extra-linguistic factors in syntactic variation and syntactic
change in our time span].
The invisible hand of grammaticalization 
which has to be studied in connection with the general development of the
Satzklammer in West-Germanic.
5. Conclusion
To sum up, grammaticalization may have therapeutic and pathological effects on
morphology. Pathological effects have been considered in this paper with regard
either to the occurrence of a morpheme in a wrong place, as is the case for introflection in Latin and “trapped” morphemes in Norwegian, or to an unexpected
form, as is the case for the perfect periphrasis in the Russian dialects and for the
West-Germanic IPP. The latter is a particularly tricky case, in which the pathological side-effect of the grammaticalization of the perfect periphrasis is responsible
for the synchronic mismatch. Because of the telic value of the prefix ge-, which was
independently grammaticalized as an additive marker in the past participle, the
AcI-verbs could not easily be target of the grammaticalization wave spreading the
perfect periphrasis throughout all the verb sets of MHG. To overcome the semantic
incompatibility, the arguably default form came in, namely the infinitive, which
assured the complete diffusion of the perfect periphrasis at the cost of a form/
meaning mismatch. A careful reconstruction based on the historical evidence available allowed us to detect the invisible but far-reaching effects of grammaticalization,
which lurk behind the synchronic mismatch. Grammaticalization qualifies as a powerful
conceptual tool for explaining quirky (morphological) features manifested by the
natural languages all over the world.
References
Abraham, Werner. 1998. “Grammatische Miszellen für Ingerid Dal”. Historische und germanische Syntax ed. by John O. Askedal, 87–132. Frankfurt/Main: Lang.
Braune, Wilhelm. 2004. Althochdeutsche Grammatik I. Laut- und Formenlehre. 15th ed. by Ingo
Reiffenstein. Tübingen: Niemeyer.
Bybee, Joan. 2006. “Language Change and Universals”. Linguistic Universals ed. by Ricardo
Mairal & Juana Gil, 179–194. Cambridge: Cambridge University Press.
Creissels, Denis. 2008. “Direct and Indirect Explanations of Typological Regularities: The Case
of Alignment Variations”. Folia Linguistica 42.1–38.
DUDEN. 2005. Die Grammatik. Band 4. 7th ed. Mannheim: Dudenverlag.
Ebert, Robert. 1986. Deutsche Syntax 1300–1750. Bern: Lang.
Ebert, Robert, Oskar Reichmann, Hans-Joachim Solms & Klaus-Peter Wegera. 1993. Frühneuhochdeutsche Grammatik. Tübingen: Niemeyer.
Eisenberg, Peter, George Smith & Oliver Teuber. 2001. “Ersatzinfinitiv und Oberfeld. Ein großes
Rätsel der deutschen Syntax”. Deutsche Sprache 29.242–260.
 Livio Gaeta
Fischer, Olga. 1994. “The Fortunes of the Latin-type Accusative and Infinitive Construction in
Dutch and English Compared”. Language Change and Language Structure. Older Germanic
Languages in Comparative Perspective ed. by Toril Swan, Endre Morck & Olaf J. Westvik,
91–133. Berlin & New York: Mouton de Gruyter.
Francis, Elaine J. & Laura A. Michaelis. 2003. “Mismatch: A Crucible for Linguistic Theory”.
Mismatch: Form-Function Incongruity and the Architecture of Grammar ed. by Elaine
J. Francis & Laura A. Michaelis, 1–27. Stanford: CSLI Publications.
Gaeta, Livio. 2005. “Ersatzinfinitiv im Deutschen: Diachrone Überlegungen zu einem synchronen
Rätsel”. Perspektiven Eins. Akten der ersten Tagung deutscher Sprachwissenschaft in Italien ed.
by Claudio Di Meola, Antonie Hornung & Lorenza Rega, 149–165. Roma: Istituto Italiano
di Studi Germanici.
Gaeta, Livio. 2008. “Mismatch: Grammar Distortion and Grammaticalization”. Rethinking Grammaticalization. New Perspectives ed. by María José López-Couso & Elena Seoane, 103–127. Amsterdam/
Philadelphia: John Benjamins.
Grimm, Jacob. 1837. Deutsche Grammatik. Vierter Teil. Göttingen: Dieterich.
Grønvik, Ottar. 1986. Über den Ursprung und die Entwicklung der aktiven Perfekt- und
Plusquamperfektkonstruktionen des Hochdeutschen. Oslo: Solum.
Härd, John E. 1981. Studien zur Struktur mehrgliedriger deutscher Nebensatzprädikate. Göteborg:
Acta Universitatis Gothoburgensis.
Harris, Alice C. 2003. “Cross-linguistic Perspectives on Syntactic Change”. The Handbook of
Historical Linguistics ed. by Brian Joseph & Richard Janda, 529–551. Oxford: Blackwell.
Harris, Alice C. & Jan Terje Faarlund. 2006. “Trapped Morphology”. Journal of Linguistics
42.289–315.
Haspelmath, Martin. 1993. “The Diachronic Externalization of Inflection”. Linguistics 31.279–309.
Hopper, Paul J. & Elizabeth C. Traugott. 2003. Grammaticalization. Second Edition. Cambridge:
Cambridge University Press.
Ijbema, Aniek. 1997. “Der IPP-Effekt im Deutschen und Niederländischen”. Groninger Arbeiten
zur germanistischen Linguistik 40.137–168.
Klein, Wolfgang. 2000. “An Analysis of the German Perfekt”. Language 76.358–382.
Kotin, Michail L. 1997. “Die analytischen Formen und Fügungen im deutschen Verbalsystem:
Herausbildung und Status (unter Berücksichtigung des Gotischen)”. Sprachwissenschaft
22.479–500.
Kotin, Michail L. 2000. “Das Partizip II in hochdeutschen periphrastischen Verbalfügungen
im 9.-15. Jh. Zur Ausbildung des analytischen Sprachbaus”. Zeitschrift für germanistische
Linguistik 28.319–345.
Kotin, Michail L. 2003. Die werden-Perspektive und die werden-Periphrase im Deutschen. Frankfurt/
Main: Lang.
Kurrelmeyer, William. 1910. “Über die Entstehung der Konstruktion ‘ich habe ihn sagen hören’ ”.
Zeitschrift für deutsche Wortforschung 12.157–173.
Lowe, Pardee. 1972. “Germanic Word Formation”. Toward a Grammar of Proto-Germanic ed. by
Frans van Coetsem & Herbert L. Kufner, 211–237. Tübingen: Niemeyer.
Paul, Hermann. 1920. Deutsche Grammatik. Band IV. Halle an der Saale: Niemeyer.
Paul, Hermann. 2007. Mittelhochdeutsche Grammatik. 25th ed. by Thomas Klein, Hans-Joachim
Solms & Klaus-Peter Wegera. Tübingen: Niemeyer.
Schmid, Tanja. 2005. Infinitival Syntax. Amsterdam & Philadelphia: John Benjamins.
Schrodt, Richard. 2004. Althochdeutsche Grammatik II. Syntax. Tübingen: Niemeyer.
The invisible hand of grammaticalization 
Solms, Hans-Joachim. 1991. “Zur Wortbildung der Verben in Hartmann von Aues ‘Iwein’ (Hs. B) und
‘Gregorius’ (Hs. A): Das Präfix ge- im System der verbalen Präfigierung. Zugleich ein Beitrag
zur Diskussion historischer Wortbildung”. Zeitschrift für deutsche Philologie 110.110–140.
Speyer, Augustin. 2001. “Ursprung und Ausbreitung der AcI-Konstruktion im Deutschen”.
Sprachwissenschaft 26.145–187.
Wischer, Ilse & Mechthild Habermann. 2004. “Der Gebrauch von Präfixverben zum Ausdruck von
Aspekt/Aktionsart im Altenglischen und Althochdeutschen”. Zeitschrift für germanistische
Linguistik 32.262–285.
Wurmbrand, Susi. 2006. “Verb Clusters, Verb Raising, and Restructuring”. The Blackwell Companion
to Syntax ed. by Martin Everaert & Henk van Riemsdijk, vol. V, 227–341. Oxford: Blackwell.